Host, Viral, and Environmental Transcriptome Profiles of the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2)

Daniel J Butler; Christopher Mozsary; Cem Meydan; David C Danko; Jonathan Foox; Joel Rosiene; Alon Shaiber; Ebrahim Afshinnekoo; Matthew MacKay; Fritz J Sedlazeck; Nikolay A Ivanov; Maria A Sierra; Diana Pohle; Michael Zeitz; Vijendra Ramlall; Undina Gisladottir; Craig D Westover; Krista Ryon; Benjamin Young; Chandrima Bhattacharya; Phyllis Ruggiero; Bradley W Langhorst; Nathan A Tanner; Justyn Gawrys; Dmitry Meleshko; Dong Xu; Jenny Xiang; Angelika Iftner; Daniela Bezdan; John Sipley; Lin Cong; Arryn Craney; Priya Velu; Ari Melnick; Iman A Hajirasouliha; Thomas Iftner; Mirella Salvatore; Massimo Loda; Lars F Westblade; Shawn Levy; Melissa Cushing; Nicholas P Tatonetti; Marcin Imielinski; Hanna Rennert; Christopher Mason

We conducted univariate analysis of the frequency differences of ACE inhibitor exposure and multivariate regression analysis to account for established risk factors for SARS-CoV-2 outcomes (i.e. age, sex, and baseline IL-6 upon admission (Zhou et al., 2020) . We use logistic regression for analysis for comparing cohort (i) as this cohort is retrospective analysis on patients with definitive outcomes. We perform survival analysis using a Cox proportional hazards model for cohort comparisons (ii) and (iii) as some patients from these cohorts are currently being treated and their outcomes are unknown (i.e., censored). We cannot determine from our data the date of infection. For the study start date for the patient, we use the date of testing positive minus seven days. Quantitative variables (i.e., age and the first IL-6 measurement) are scaled to [0,1] to facilitate comparison of model coefficients to those of dichotomous variables. Prior to conducting our multivariate analysis, we evaluated and removed correlated covariates. We include a model built with all covariates as well in the supplemental tables.